Table 1.
Examples of collaborative studies
| Subject/collaboration | Year established | Number of studiesa | Number of individualsa | Collaborative model | Considerations | Website |
|---|---|---|---|---|---|---|
| Cardiovascular | ||||||
| Asia Pacific Cohort Studies Collaboration (APCSC)74 | ∼1999 | 44 | 600 000+ | Data are not available for use by researchers outside of the APCSC, and researchers within the APCSC wishing to work on the data have to do so in Sydney, with prior approval from the Executive Committee | Restricted to large cohort studies; broad scope of outcomes, yet not all cohorts collect all necessary data and no standardization for what is collected | http://www.apcsc.info/ |
| Cancer | ||||||
| Vitamin D Pooling Project (VDPP)75 | 2007 | 10 | 4539 cases | A subset of cohorts from the National Cancer Institute Cohort Consortium; nested case–control study using a central laboratory and standardized testing protocols | Formed with a very specific goal: to address the gap in knowledge about association between vitamin D and rarer cancer sites | https://epi.grants.cancer.gov/VitaminD/ |
| Collaborative Group on Hormonal Factors in Breast Cancer42 | ∼1996 | ∼54 | 147 000+ | All studies that included ≥100 women with breast cancer with information on the use of hormonal contraceptives and reproductive history | Cohort studies were included using a nested case–control design, matching four random controls to each case | NA |
| Breast and Prostate Cancer Cohort Consortium (BPC3)76 | 2003 | 10 | 30 000+ | A subset of cohorts from the National Cancer Institute Cohort Consortium; case–control study where cases came from member cohorts and consortium funding helped enrol additional cases/controls | Case–control study to identify genetic variants associated with cancer risk | https://epi.grants.cancer.gov/BPC3/ |
| Pooling Project of Prospective Studies of Diet and Cancer77 | 1991 | 16 | 1 026 547 | Summary data from prospective cohorts with validation data on dietary assessment data analysed centrally using two-stage analysis (random effect for cohort), standardized for all analyses | Comprehensive analysis plan published in American Journal of Epidemiology (2006) | https://www.hsph.harvard.edu/pooling-project/ |
| Child health | ||||||
| Environmental influences on Child Health Outcomes (ECHO)47 | 2016 | 84 cohorts funded in 35 awards by NIH | ∼50 000 | Diverse set of existing cohorts will harmonize extant data and agree to prospective collection of new data elements; participating cohorts have access to extensive laboratory support and expertise in measuring participants reported outcomes; research primarily structured around key outcomes—pre-, peri- and post-natal outcomes, airway conditions and sleep, neurodevelopment, obesity and positive health | ECHO collaboration also includes Institutional Development Award (IDeA) States Pediatric Clinical Trials Network, ideally to be able to test interventions that are hypothesized to improve child health based on observational study findings in medically underserved and rural populations | http://www.echochildren.org/ |
| Diabetes | ||||||
| EURODIAB ACE78 | 1993 | ∼24 | ∼30 million | Arose out of a prior collaboration established in 1985 (EURODIAB); collaborations monitored population incidence rates of childhood diabetes across Europe | Main outcome was rate of diabetes, which was susceptible to differential ascertainment; all study sites applied capture-recapture methods to monitor and report ascertainment completeness | |
| Genetics | ||||||
| Cohorts for Heart and Aging Research in Genomic Epidemiology (CHARGE) Consortium4 | 2008 | 5(+) | 38 000 | Series of prospectively planned joint meta-analyses of GWAS; studies opt in to analyses and cohorts beyond the five official members are often invited as co-collaborators (co-collaborators often collaborated with CHARGE cohorts prior to the formation of CHARGE) | Studies used different genotyping platforms that resulted in <60 000 SNPs in common; need for phenotype standardization across cohorts; scope of collaboration exceeded scope of individual studies and stretches limited resources (e.g. need to collect phenotype data beyond what was originally proposed); within-study analysis followed by between-study meta-analysis avoids need for individual-level data sharing | http://www.chargeconsortium.com/ |
| HIV | ||||||
| Antiretroviral Therapy Collaborative Cohort (ART-CC)79 | 2000 | 19 | 12 574–74 000+ | Cohorts submit data according to a protocol that standardizes variable definitions, formats and coding; all analyses are approved by a steering committee | Lots of heterogeneity in patient characteristics (% female, % IDU, AIDS, CD4+ cell count at baseline) and attrition rates (2–18%) | http://www.bristol.ac.uk/art-cc/ |
| Concerted Action on SeroConversion on AIDS and Death in Europe (CASCADE) Collaboration80 | 1997 | 28 | 25 000+ | A cohort of seroconverters (subgroups of existing cohorts); anonymized data are collected; all analyses are approved by a steering committee | Identification of seroconverters and estimated date of seroconversion varies across studies; collaborative is part of EuroCoord, a ‘Network of Excellence’ that provides scientific oversight for CASCADE and other large cohorts and collaborations (www.eurocoord.net) | http://www.ctu.mrc.ac.uk/our_research/research_areas/hiv/studies/cascade/ |
| Centers for AIDS Research Network of Integrated Clinical Systems (CNICS)38 | 2006 | 8 | 30 000+ | Data extraction of a pre-defined set of variables; anonymized prior to submission to Data Coordinating Site; all analyses are approved by a steering committee; individual cohorts opt in to participate in any proposed analysis | Includes historical data on patients in care from 1995+ for some sites; collaboration also coordinates research on stored specimen maintained in repositories at each individual study site | https://www.uab.edu/cnics/ |
| Collaboration of Observational HIV Epidemiological Research Europe (COHERE)81 | 2005 | 40 | 300 000+ | Uses HIV Cohorts Data Exchange Protocol82—a standardized method of data structure and transfer—to compile data | Includes other, smaller collaboratives within it—e.g. CASCADE, EuroSIDA, UK CHIC | http://www.cohere.org/ |
| HIV Cohorts Analyzed Using Structural Approaches to Longitudinal Data (HIV-CAUSAL)83 | ∼2010 | 12 | 63 000+ | Originally formed to answer: When to start ART? What regime to start? And when to switch regimes? | Explicitly created to be analysed using causal methods; includes historical data on patients in care from 1996 to 1998+ for all cohorts | https://www.hsph.harvard.edu/miguel-hernan/hiv-causal-collaboration/ |
| International Epidemiologic Databases to Evaluate AIDS (IeDEA)2,84,85 | 2005 | >50 | 1 700 000+ | This is a collaboration of regional collaborations across the world, encompassing North America, Central and South America, Asia and Pacific Islands, and three African regions;7 analyses must be approved by regional steering committees or, to use data from entire collaborative, by an executive committee | Scale of heterogeneity increased due to increased number of sites; includes other collaboratives (sometimes collaboratives of collaboratives) within it—e.g. NA-ACCORD is the North American regional representative | http://www.iedea.org/ |
| Kidney disease | ||||||
| Chronic Kidney Disease Prognosis Consortium86 | 2009 | 46 | 2 100 000+ | Cohorts can share either individual de-identified data or can run a standard programme to create all output needed to include their study in meta-analytic results | Restricted to prospective studies with ≥1000 participants (except CKD cohorts) and ≥1 outcome of interest with a minimum of 50 events | http://www.jhsph.edu/research/centers-and-institutes/chronic-kidney-disease-prognosis-consortium/ |
a
Most recent or approximate estimate; most cohorts are open and continue to enrol members and most collaboratives continue to enrol (and lose) individual studies.
CKD, Chronic Kidney Disease; GWAS, Genome-Wide Association Studies.